Long ago I thought of building a circuit that does this:

- takes 2 inputs, "low" and "high", signifying low fan speed, and high fan speed
- has 2 outputs for 2 fans
- when "high" input is active, both fans receive full battery voltage
- when "low" input is active, the fans are placed in series so they both run at low speed.

When one has a common shroud with 2 fans (as opposed to 2 independent shrouds, one per fan), running both fans at low speed is better than having one fan running, because in that scenario, some of the air goes forwards through the non-running fan's blades, from the engine bay, and out the running fan - iow some air is bypassed around the radiator, making it inefficient.

I'm talking about this shroud style:


Additionally the fans will last longer as running at reduced speed does that. And less noisy if you care lol.

Additional features of circuit:
- if both "low" and "high" are active, "low" will be ignored and the fans will receive full voltage.

- when going from "off" to "high" or "low" to "high" the 2 fans will have a staggered turn on - the 2nd will have a delayed turn on (1 or 2 seconds) so as to reduce the idle dip.


With a programmable ECU you would set it up to say go "low" at 95*C and "high" at 98*C, with 2* of hysteresis. (If you have a t-stat <92*C). A/C turns both fans on.


The circuit will have to have 3 relays on board.


I've thought of building such a circuit, but just wired my fans together lol.

If anyone wants to build and test this circuit idea, I can draw up a schematic......

 

Id like to run something like this, my mx3 fans are on their way out and need replaced anyway

 

I've thought about doing something different: replacing the relays with DC PWM drivers. So when the circuit that replaces the relay gets a "switch on" signal, it will slowly ramp up from the start duty cycle to the max duty cycle. This will also prevent an idle dip.

Obviously you would need a powerful driver to do this, and one heck of a heatsink. So that takes away the possibility of putting it inside an actual gutted relay.

 

The 3rd Gen RX-7, for one, used exactly the type of control as what you're describing. In this day it's needlessly complex IMO. If your ECU has PWM based control you can use a solid state relay. Setup this way the fan will turn just fast enough to keep things within the programmed temp range. Many OEs have gone this route and many aftermarket ECUs support this strategy.

 

If you are talking about the FD, then it doesn't have any PWM control, instead it has two fans and each fan has 3 different speeds, 4 relays in total, but they always work in parallel IIRC. They are control both by the ECU as well as by a thermoswitch on the pump.

The FD's complexity is something that should be avoided at all costs.


And btw, solid state relays are hugely expensive. Like $60 for a 10A relay.

 

And closer to $100 for a 30A. I've looked into it before--too expensive.

Jason, if you want to draw it up, I can put it on the test bench.

 

Don't OEM's already use these? First one that comes to mind is the one on the early EVO's that they recalled. Mitsubishi P#1355A053?

EDIT: I just flipped through the EVO VIII wiring diagram, and that P# appears to be a single channel speed controller. The A/C fan runs off a simple relay. :(

 

I know that the 99+ jeep grand cherokees use a solid state relay for fan control. I've seen one get replaced for about $40. Probably worth looking into.

There's some discussion in the ms3 dev forum about using a FET to do it, would be quite a bit cheaper and I think they found a few parts that should work. I don't think a heat sink would necessarily be a deal breaker for everyone.

I had a 'delta current control' fan controller on my v8 rx7, and that thing was win. If anyone doesn't know the benefits of what we're talking about you should read about it.

http://www.dccontrol.com/

 

I ran one of those on my Lightning and it was flawless.

 

I can use MOSFETs but I thought the heatsinking requirement would be undesireable.

Infineon make D2PAK (surface mount) 1 mΩ MOSFETs - with a 20A load they will dissipate over 0.4W, which is manageable using PCB copper area heatsinking. However the usual suspects, Mouser and Digikey, don't stock 'em.

There are 1.7~2 mΩ MOSFETs available, some in TO-220 (non-SMD), but they will then dissipate ~2x the power, and TO-220 parts will require a real heatsink. I can parallel them up though. That will reduce the total power dissipation as well as spread the heat out.

If I will go through the trouble of using MOSFETs, then I can PWM em as mentioned, and soft start them on... and just power the fans in parallel. And have them run variable speed, circa 6V at 95*C, sliding to full voltage at 99*C or something like that.
Hmmm...

If so, would it take a 0-5V signal from the ECU signifying fan speed?
Do most ECU's have a spare PWM output to generate a 0-5V based on some whatever signal? I'd say no..

So would a simple "low and high" with soft-start circuit be the most useful? (with a single output wherein the fans are paralleled)

 

My '96 Talon had two speed fans as a factory setup. I don't know how the circuit looks but you might do some digging around and see if you can find some info. I'll try to find a diagram somewhere in my pile of DSM info.

 

My M3's fan has 2 inputs, hi and low. I spose the motor has 2 sets of windings.

 

If you can design the analog part, I can design the digital part...

 

Exactly. Similar to what the original poster was suggesting.

Also, Hella offers a 32A solid state relay that is available for around $40.

 

Dimitris, I know you can even do it with the main processor.

I'm thinking for maximum usefulness across a broad range of applications a simple low and a high input and a single, paralleled fan output would be good. This would then have a soft start function, and PWM for the 6V output / low speed state. Use MOSFETs and no relays. Current limit and UVLO would be good too.

At this point a small 8-pin PIC processor would yield the most compact design.
I can design the MOSFET and current sense circuits.

 

I'm not using PICs but Atmel AVRs, and I'm sticking to them as I have all the hardware and software required for AVRs. Do you have any specific reason as to why you want to go with a PIC?

 

Only bec it was the first thing that came to mind when I thought "8 pins".

 

In which case, we can use a ATTiny13.

 

I thought there would be an 8-pin AVR.
So do you wanna do it?
My circuit would have a bunch of MOSFETs, and a current sense resistor and amplifier or comparator to shut it down or current limit on a short circuit.

 

The ATTiny13 *is* an 8-pin AVR. Why use a bunch of MOSFETs? Cut the thermal load in each by half? A current sense resistor is not a problem, we can use a 1mOhm shunt resistor or go serious and use the Allegro ACS715 which can measure up to 30 amps.

So:

Out of the 8 pins, the 6 are usable:
- One pin for the start/stop command (pull the pin low to start)
- One pin for the PWM
- One pin for the current sensor
- One pin to a LED to signify power on
- One pin to a LED to signify that the fan should be working (ie DC > 30%).
- One pin to signify an error status (overcurrent condition?)

This could be a drop in replacement for any fan relay, the 4 pins of the relay are exactly what is needed to get it to work.